THE CHARACTERISTICS AND TRADING BEHAVIOUR OF
DUAL-LISTED COMPANIES
Jaideep Bedi, Anthony Richards and Paul Tennant
Research Discussion Paper
2003-06
June 2003
International Department
Reserve Bank of Australia
We thank Chris Becker for extensive helpful discussions, and Ric Battellino,
Jonathan Kearns and Ellis Connolly for comments. The views expressed in this
paper are those of the authors and do not necessarily reflect the views of the
Reserve Bank of Australia.
Abstract
We examine the characteristics and stock price behaviour of existing and recently
unified dual-listed companies (DLCs, also known as Siamese-twin companies).
DLC structures are effectively mergers in which companies agree to combine their
operations and cash flows, but retain separate identities and shareholder registries.
We identify 14 such international structures and survey the rationales that have
been advanced for the creation as well as the unification of such groups.
We find that three recent Anglo-Australian DLCs exhibit the ‘excess comovement’
phenomenon identified by Froot and Dabora (1999) and confirm this phenomenon
has persisted for the long-standing Anglo-Dutch DLCs. We also investigate what
happens to the market exposures of DLCs that have been abandoned in favour of a
unified structure. Standard models would suggest there should be no change in the
betas of the combined firm, while models of trading-based comovement would
suggest that betas could change. We find that the market value of the unified DLCs
becomes more (less) correlated with the market index of the new primary
(secondary) market after unification. Together with the evidence for excess
comovement, this result is consistent with a model where the market prices of
assets depend not only on fundamentals, but also on the location of trade and the
investors that hold the assets.
Finally, we conduct an event study into the stock returns of DLC twins around the
time of unification announcements. Unifications of the share structure have
typically occurred on the market that placed the higher value on the cash flows of
the DLC. Not surprisingly, the pricing of the twins converges after these
announcements, and we find that a rise in the value of the discounted twin is
apparently accompanied by a modest fall in the value of the twin trading at a
premium.
JEL Classification Numbers: F30, G12, G34
Keywords: dual-listed company, Siamese twin companies, international equities
i
Table of Contents
1.
Introduction
1
2.
Why Do Companies Choose DLC Structures?
4
3.
Examples of Dual-listed Companies and Puzzles in Their Pricing
8
4.
Data
17
5.
Testing for Excess Comovement
17
6.
Testing for Changes in Market Exposures Following Unification
of DLCs
23
7.
An Event Study of Announcements of DLC Unifications
29
8.
Conclusion
34
References
36
ii
THE CHARACTERISTICS AND TRADING BEHAVIOUR OF
DUAL-LISTED COMPANIES
Jaideep Bedi, Anthony Richards and Paul Tennant
1.
Introduction
The phenomenon of substantial price differences between dual-listed (or ‘Siamese
twin’) companies has been very widely cited following the influential work of
Froot and Dabora (1999). Dual-listed company (DLC) structures are effectively
mergers between two companies in which they agree to combine their operations
and cash flows and make similar dividend payments to shareholders in both
companies, while retaining separate shareholder registries and identities. In this
respect a dual listing is quite different to a cross listing. Whereas a dual listing
involves the quasi merger of two separate entities, a cross listing occurs when an
individual company establishes a secondary listing on a foreign stock exchange,
the most prominent arrangement being via American Depositary Receipts
(ADRs).1
Apart from the work of Froot and Dabora and Rosenthal and Young (1990), there
appears to be no other work looking at this interesting phenomenon. Yet DLC
structures are not an archaic structure, and this paper documents 12 DLC structures
put in place over the last 15 years, in addition to 2 longstanding cases (Royal
Dutch/Shell and Unilever) made prominent by these earlier papers. Indeed, in the
recent battle to take over UK-listed P&O Princess Cruises Plc, both US suitors
(Royal Caribbean Cruises Ltd and Carnival Corporation) proposed to merge via a
DLC structure. The rationale for DLC structures appears to have been that the
structure has been a key element for some companies in facilitating cross-border
mergers or international expansion. However, many of the recently formed DLCs
have not proved durable, with 6 of the 12 recent DLCs already unified into a more
conventional single share structure.
1
Karolyi (1998) provides a survey discussion of cross listings and Pagano, Röell and Zechner
(2002) examine the motives for cross listings.
2
DLCs are an interesting phenomenon because of the insights they can yield about
both corporate structure and asset pricing. Although the shares of the two
companies represent identical claims on the future cash flows of the group,
substantial divergences in the pricing of these claims are observed. Rosenthal and
Young document the persistent mispricing of Royal Dutch/Shell and Unilever
throughout the 1980s but find that transactions costs would have precluded trading
rules designed to take advantage of these price differentials. Froot and
Dabora (1999) investigate whether stock returns of DLCs are affected by the
location of trade. They find that the return of each ‘twin’ appears to be correlated
with the market on which it is most actively traded. For example, when the New
York market rose relative to the UK market, they found that the price of Royal
Dutch (which traded relatively more in New York over their sample) increased
compared to the price of Shell (which traded relatively more in London).
Accordingly, the mispricing of DLCs is frequently cited as representing an
anomaly to the efficient market hypothesis (Mullainathan and Thaler 2000;
Barberis and Thaler 2002). In particular, Froot and Dabora’s evidence is
interpreted as evidence that stock prices reflect in part the sentiment and
behavioural biases of investors and market-makers that dominate trading in the
asset in question. The findings are not unique in this respect. A related example is
the case of closed-end mutual funds which often trade at substantial premia or
discounts to their net asset value, a result that Lee, Shleifer and Thaler (1991) and
Shleifer (2000) attribute to small-investor sentiment.
This paper presents new evidence into four aspects of DLCs. First, since DLCs can
now be said to be a reasonably widely used corporate structure, we document the
formation of 12 new DLC structures over the last 15 years and outline the reasons
that have been cited for the adoption of such structures, and for the subsequent
unification of the share structure where this has occurred.
Second, we show that excess comovement and substantial price divergences
between DLC twins have remained pervasive phenomena in six large remaining
DLCs. If the large price divergences identified by Rosenthal and Young and Froot
and Dabora did represent a source of arbitrage profit, we would expect arbitrage
activity to have occurred and resulted in a reduction in the size of the typical price
gap. We note several instances where particular institutions have attempted to find
3
ways to arbitrage the price differences, but conclude that the ongoing price
differences suggest little success in this regard.
Third, we study what happens to the betas or market exposures of DLC companies
in cases where the structure is abandoned in favour of a single entity with one
primary listing. In particular, we identify four cases where such unifications were
not accompanied by other changes in the underlying business of the companies and
examine if the betas of the combined company change as a result of it changing
from a dual listing on two exchanges to a single primary listing on one exchange.
A traditional view of asset pricing would suggest there should be no change in the
betas of the combined company following such a change. By contrast, the model of
trading-induced comovement proposed by Barberis, Shleifer and Wurgler (2002)
would suggest an increase in the beta with respect to the market which is the new
primary listing of the unified company, and a decrease in the other market where
there is no longer a primary listing. Since the unifications result in compositional
changes in the international equity indices produced by major index providers –
there is an increase in the stock’s index weight in the former market and a deletion
from the index of the latter market – our test is related to the tests of Barberis,
Shleifer and Wurgler (2002) who examine changes in betas following inclusions or
exclusions from the US S&P 500 index. Our results suggest that the betas do
indeed change in the manner predicted by a model of trading-induced
comovement. Our results are also consistent with the results of Chan, Hameed and
Lau (2003) who show that the exposures of four prominent Hong Kong companies
in the Jardine Group experienced large changes in their market exposures
following the change of their primary listing to the Singapore Exchange. Together
with the earlier evidence for excess comovement in DLCs, this result is consistent
with a range of related evidence that the market prices of assets depend not only on
fundamentals, but also on the location of trade and the investors that hold the
assets.
Finally, we conduct an event study into the behaviour of the market value of DLCs
in cases where the share structure is unified. Not surprisingly, the pricing of the
twins converges following the announcement of unification, but it is of interest to
ask if this occurs via an increase in the share price of the company that is trading at
a discount or a fall in the share price of the twin that is trading at a premium, and if
there is any impact on overall company value. The tests suggest a rise in the price
of the twin trading at a discount, with some modest evidence of a fall in the
4
company trading at a premium. On average, there is little change in total firm
value, implying that we can say little from this event study about the way that
markets view DLC structures versus conventional mergers.
The rest of this paper is arranged as follows. Section 2 provides an introduction to
DLCs, a listing of DLC structures, and discusses some of the reasons cited for and
against such a structure. Section 3 documents the price differentials existing in the
cases of the six large continuing DLCs, and discusses some of the reasons why
they are not eliminated. Section 4 outlines the data used in the three empirical
sections of the paper. Sections 5, 6, and 7 contain the results from the tests for
excess comovement and for changes in market exposures following the unification
of DLCs, and from the event study for changes in market values following
unification announcements. Our conclusions are presented in Section 8.
2.
Why Do Companies Choose DLC Structures?
DLC structures are effectively mergers between two companies in which they
agree to combine their operations and cash flows, but retain separate shareholder
registries and identities. One form of DLC involves the two companies transferring
their assets to one or more jointly owned subsidiary holding companies. The
holding company then passes dividends back to the main companies, which
distribute them according to a predetermined ratio. Alternatively, instead of the
transfer of assets, there may be contractual arrangements to share the cash flows
from each other’s assets. The operations of the two companies are closely
coordinated, and in most cases the companies share a common board of directors.
5
Table 1: Dual-listed Companies
Company
Country
Period of DLC
Shell Transport & Trading Co PLC
Royal Dutch Petroleum
Unilever PLC
Unilever NV
ABB AB
ABB AG
SmithKline Beecham PLC
SmithKline Beecham
Fortis (B)
Fortis (NL)
Reed Elsevier PLC
Reed Elsevier NV
Rio Tinto Limited
Rio Tinto PLC
Dexia Belgium
Dexia France
Nordbanken
Merita
Allied Zurich PLC
Zurich Allied
BHP Billiton Limited
BHP Billiton PLC
Brambles Industries Limited
Brambles Industries PLC
Investec Limited
Investec PLC
P&O Princess Cruises PLC
Carnival Corporation
UK
Netherlands
UK
Netherlands
Sweden
Switzerland
UK
US
Belgium
Netherlands
UK
Netherlands
Australia
UK
Belgium
France
Sweden
Finland
UK
Switzerland
Australia
UK
Australia
UK
South Africa
UK
UK
US
Since 1907
Since 1930
January 1988–July 1999
July 1989–April 1996
June 1990–December 2001
Since January 1993
Since December 1995
November 1996–February 2000
December 1997–March 2000
September 1998–October 2000
Since June 2001
Since August 2001
Since July 2002
Since April 2003
Table 1 provides a listing of 14 existing or recently unified DLC structures.2 With
one exception, all DLCs have been the result of mergers between companies
2
The table includes all those widely held DLC structures over recent decades that could be
identified from a range of sources. It excludes cases of twins that do not trade separately. For
example, in the Anglo-Irish Wedgwood/Waterford merger, shareholders in each company
received an equity unit that consists of a share in each company. A similar arrangement
occurred in the creation of the Anglo-French EuroTunnel enterprise. Unlike the cases
discussed in this paper, the shares of the companies do not trade as different companies as the
equity units cannot be split. The table also excludes cases of linked companies that did not
have identical dividend flows.
6
domiciled in different countries.3 An examination of these cases suggests that
companies may choose DLC structures rather than conventional mergers for a
number of reasons:4
•
Tax or accounting factors. A DLC structure may minimise capital gains tax
obligations that would result from a conventional merger. Alternatively,
home-bias and differences in national tax systems may favour a DLC
structure whereby cross-border dividend payments to shareholders are
minimised. Similarly, accounting regimes may favour a DLC over a
conventional merger or acquisition if the latter would require recognising and
amortising goodwill that results from the merger.
•
National identity issues and foreign investment regimes. A conventional
merger or takeover would result in the disappearance of one of the companies.
Since complicated cross-border mergers typically require various forms of
official approval, DLCs that preserve the existence of each company in each
market may be the best way of ensuring that approval. In addition, in cases
where the two companies are of similar size, the management of the
companies may both wish to avoid the appearance of having been taken over.
•
Operational and corporate governance issues. The existing contractual
arrangements of the companies may cause various types of rights to be
triggered (e.g., options in debt contracts, and rights of other companies
involved in joint ventures) in the event of a takeover or conventional merger.
3
The exception is the recent case of Investec PLC/Limited. This Anglo/South African DLC
was formed not from a merger, but from a ‘demerger’ and the creation of a new UK company
holding the UK assets of the South African parent. As with other DLCs, the purpose of the
transaction was to facilitate Investec’s international expansion. However, the rationale for a
DLC structure rather than a simple UK secondary market listing appears to have been to meet
South African government requirements concerning exchange controls. Since Investec has a
substantially smaller capitalisation than the other DLCs, this article focuses on the more
conventional and larger DLCs created through mergers.
4 Some of the reasons cited by companies in adopting DLCs are provided in the 12 April 2001
‘Proposed DLC Merger Explanatory Memorandum’ from BHP Billiton and the 25 June 2001
‘Information Memorandum: Dual Listed Companies Proposal’ from Brambles.
Hancock, Phillips and Gray (1999), Glanz and Sanderson (2001), Smith and Cugati (2001),
and Hancock, Gray and Sommelet (2002), also provide further discussion of DLCs, and their
advantages and disadvantages relative to conventional mergers.
7
However, these consequences may be avoided if the merger occurs in the
form of a DLC arrangement.
•
Perceptions of better access to capital markets. Since local investors are
already familiar with their respective companies, management may believe
that the merged company will have better access to capital markets if it
maintains listings in each market.
•
Concerns over ‘flow-back’. In a conventional merger with a stock swap, the
merged company will have to choose one country for its domicile and primary
listing, and the shareholders from the other country will receive equity in a
company domiciled in a foreign market. The merged company will now be a
larger company and will see a higher weight in the share market index of its
country of domicile, but it will disappear from the index in the other market.
A DLC may be chosen if it is thought that a merger would result in selling
pressure in one market exceeding increased investor interest in the other
market.
However, the fact that most cross-border mergers do not take the form of a DLC
and that some companies have decided to unify their DLC structures implies that
there are also possible disadvantages to DLCs. These may include:5
•
Complexity of operations. The contractual arrangements of DLCs provide
procedures for the treatment of the interests of the shareholders of both
companies in the case of capital raisings, asset sales and other events.
Nonetheless, the existence of two sets of shareholders may at times constrain
the flexibility of management and the full benefits of a more conventional
merger may not in practice be realised.
•
Regulatory issues. The ongoing operations of the separate companies means
that the DLC must satisfy the accounting and regulatory frameworks of two
countries. This is likely to be costly, and may constrain the ability of
management to maximise the joint value of the two companies.
5
Further details on the factors that lead companies to end DLC structures are provided in the
28 August 2000 press release by Fortis, the 17 April 2000 ‘Share unification plan’
presentation of Zurich Financial Services, and the 1999 Annual Reports of Dexia, ABB and
MeritaNordbanken.
8
•
3.
Liquidity, transparency and shareholder value issues. In practice, the
existence of two separate companies may result in less share market liquidity
than would result if there was a single larger company. In addition, investors
may view the DLC structure as somewhat complex and less transparent than a
conventional single company. Hence, they may value the two parts of the
company less highly than they would a single larger company.
Examples of Dual-listed Companies and Puzzles in Their
Pricing
Of the six large DLCs existing in early 2003, three involved companies from the
United Kingdom and the Netherlands (Royal Dutch Shell, Unilever and Reed
Elsevier) and three involved companies from the United Kingdom and Australia
(Rio Tinto, BHP Billiton, and Brambles Industries). We refer henceforth to these
as the Anglo-Dutch and Anglo-Australian DLCs.
The concentration of UK, Australian and Dutch companies in this group raises the
question of whether there is something peculiar to these countries to encourage
DLCs. No obvious explanation appears to exist, though one possible factor may be
that a listing on the London equity market is viewed as particularly attractive, and
that given the choice between a single listing on another market and a dual listing
that also includes London, companies may choose not to give up their London
listing.
Given that Froot and Dabora concentrated on two of the Anglo-Dutch cases, we
begin our discussion of the continuing DLCs with a presentation of some basic
facts about the more recent Anglo-Australian DLCs. The first of these resulted
from the 1995 merger of Australian mining company CRA and UK-listed RTZ,
which already held a 49 per cent stake in CRA. The two companies have
subsequently been renamed Rio Tinto Limited (which is traded on the Australian
Stock Exchange) and Rio Tinto PLC (which is traded on the London Stock
9
Exchange). The sharing agreement stipulates that the dividend and capital rights of
each PLC share relative to each Limited share are on a 1:1 basis.6
Given that the two classes of Rio Tinto shares entitle the holder to exactly the same
flow of dividends, one might have expected that they should have traded at the
exact same price. An examination of the share prices shown in Figure 1 shows that
they are highly correlated.7 However, significant and persistent price divergences
have existed, even after using a 20-day moving average to mitigate any impacts
from asynchronous trading. Each company has traded at a significant premium at
particular times, and the average of the price differential over the full seven-year
period has been a premium of the UK listing over the Australian listing of around
2 per cent.
BHP, the large diversified resource company, was the second Australian company
to enter into a DLC structure when it merged with Billiton, a UK/South African
resources company in July 2001. The equalisation ratio is 1:1, such that a UK share
has the same economic value and voting rights of an Australian share.8 From the
outset, the relative value of the Australian scrip has been higher than its UK
counterpart (Figure 2) with the premium averaging 8 per cent.
6
The RTZ holding in CRA has since been reduced to less than 40 per cent. Under the terms of
the merger, in cases where the companies did not share a common interest, the RTZ stake in
CRA was not to be used to cast votes, implying that PLC (Limited) shareholders hold 76.5
(23.5) per cent of votes on joint decisions.
7 Indeed, the standard tests suggest that the two price series for each of the three AngloAustralian DLCs and Anglo-Dutch DLCs (when expressed in a common currency) are
cointegrated. Data sources for the stock prices and exchange rates used in this section are
provided in Section 4.
8 Shareholders in the Australian (UK) arm hold 60 (40) per cent of the combined company.
10
Figure 1: Rio Tinto Limited and PLC
Share prices and price differential
A$
A$
Rio Tinto PLC
35
35
25
25
Rio Tinto Limited
%
5
%
5
0
0
-5
-5
-10
-15
-10
Premium to Limited (4-week moving average)
l
l
1996
l
1997
l
1998
l
1999
2000
l
2001
-15
2002
Figure 2: BHP Billiton Limited and PLC
Share prices and price differential
A$
A$
BHP Billiton Limited
11
11
10
10
9
9
BHP Billiton PLC
8
8
%
%
Premium to Limited
(4-week moving average)
10
10
5
5
0
l
l
J
l
l
l
A
O
2001
l
l
l
D
l
l
F
l
l
A
l
l
l
J
A
2002
l
l
l
O
l
D
0
11
The most recent Anglo-Australian DLC was formed in August 2001 when
Brambles, the Australian industrial firm, merged with the industrial services arm of
UK-listed GKN. Just as there had been pre-existing ownership links between the
Rio Tinto companies, Brambles and GKN had an existing relationship, including
joint ventures. Similar to the other two Australian DLCs, the equalisation ratio is
1:1.9 Figure 3 shows the sizeable premium, averaging 9 per cent, at which the
Australian share has traded relative to the UK share.
Figure 3: Brambles Limited and PLC
Share prices and price differential
A$
A$
Brambles Industries Limited
10
10
8
8
Brambles Industries PLC
6
6
4
4
%
%
Premium to Limited
(4-week moving average)
10
10
5
5
0
l
l
O
2001
l
l
D
l
l
F
l
l
A
l
l
J
2002
l
l
A
l
l
O
l
D
0
The price differentials of the three Anglo-Australian DLCs are shown in Figure 4.
The premia have at times moved together, however, there have also been times
when they have moved in opposing directions. Although the premium for
Rio Tinto has changed sign over the seven-year period of the DLC, in the cases of
BHP Billiton and Brambles (for which only around 18 months of data exist), the
Australian twin has consistently traded at a premium. Data for the median absolute
price differentials are shown in Table 2.
9
Shareholders in the Australian (UK) arm hold 57 (43) per cent voting interest in the combined
group.
12
Figure 4: Price Differentials for Anglo-Australian DLCs
4-week moving average of daily premia
%
%
Australian firm at a premium
Brambles Industries
10
10
BHP Billiton
5
5
0
0
-5
-5
Rio Tinto
-10
-10
Australian firm at a discount
-15
l
1996
l
1997
l
1998
l
1999
l
2000
l
2001
2002
-15
Royal Dutch Petroleum and Shell, Unilever NV/PLC, and Reed Elsevier NV/PLC
are the three other examples of continuing DLCs on overseas markets.10 Created in
1907, Royal Dutch/Shell is the oldest example of a DLC. Both companies trade on
a number of exchanges, but Royal Dutch is incorporated in the Netherlands while
Shell is incorporated in the United Kingdom. The Unilever group was formed in
1929 from the merger of a Dutch margarine company, Margarine Unie and Lever
Brothers, a British soap maker. Unilever NV and Unilever PLC are listed on the
Dutch and UK stock exchanges respectively. Finally, Reed Elsevier was formed
from the January 1993 merger of Reed International PLC, a UK-listed publishing
and information company, and Elsevier NV, a Dutch-listed publishing company.
10
The groups’ interests are split respectively as follows: 60:40 in favour of Royal Dutch; 50:50
for Unilever; and 52.9:47.1 in favour of Reed Elsevier PLC.
13
Table 2: Price Differential for Anglo-Dutch and Anglo-Australian DLCs
Mean absolute deviation – per cent
Royal Dutch/Shell
Unilever
Reed Elsevier
Rio Tinto
BHP Billiton
Brambles Industries
Notes:
1980–1995
1996–2002
September 1999–2002
9.53
8.86
–
–
–
–
8.48
9.59
9.28
3.92
–
–
4.70
8.06
8.20
4.36
7.39
8.87
This table shows the mean absolute deviation from price parity (in per cent) for each of the six DLC
arrangements, calculated using end-week data. The period 1980–1995 corresponds to the sample period
in Froot and Dabora (1999), and 1996–2002 corresponds to the period following their sample period. The
period September 1999–December 2002 corresponds to the period following the publication of the article
by Froot and Dabora (1999) in the Journal of Financial Economics. The data for BHP begin in
April 2001 and the data for Brambles Industries begin in August 2001.
The long-run behaviour of the price differentials for the three Anglo-Dutch DLCs
is shown in Figure 5. As has been documented by Froot and Dabora (1999), the
price differentials for Royal Dutch/Shell and Unilever are surprisingly large and
variable. It is, however, noteworthy that the extreme price differentials seen in the
early 1980s have not been repeated, which may reflect the greater integration of
capital markets and the reduction of transactions costs over the last two decades.
Interestingly, there is some evidence of correlation across the three DLCs. At one
level this might not appear entirely surprising if the excess comovement result of
Froot and Dabora is an extremely persistent phenomenon rather than just a shortrun one. However, the fact that there is substantial variance in the proportion of
trading of the Dutch twins that occurs on different markets implies that the
correlation might not be so easily attributed to common influences from the
relative performance of national markets.11
11
In particular, a substantial fraction of trade in Royal Dutch occurs in the US market, whereas
almost none of the trade in Reed Elsevier NV occurs in that market.
14
Figure 5: Price Differentials for Anglo-Dutch DLCs
4-week moving average of daily premia
%
UK firm at a premium
%
40
40
Royal Dutch/Shell
30
30
20
20
Reed Elsevier
10
10
0
0
-10
-10
Unilever
-20
-20
UK firm at a discount
-30
l
l
l
1982
l
l
l
l
1986
l
l
l
l
1990
l
l
l
l
1994
l
l
l
l
1998
l
l
l
-30
2002
If there is a tendency for anomalies documented by researchers to be reduced or
eliminated after their publication, one might have expected to see a decline in the
magnitude of DLC price differentials following the publication of Froot and
Dabora’s work. Although the sample is too short to draw firm conclusions, the data
for median absolute price differentials in Table 2 appear to provide only limited
evidence for this. (Further, the evidence would be weaker if the comparison instead
used the date of the publication of the working paper version of the article). In
particular, although the mean absolute deviation for Shell has fallen substantially
relative to historical levels, the declines in the other two cases are much less
obvious.
Just as Froot and Dabora were unable to identify ‘fundamental’ factors that could
explain the price differentials that they identified in the Anglo-Dutch cases, so too
do fundamentals-based explanations appear unable to explain substantial price
differences between the Australian and UK arms of the three more recent DLCs.
For example, the stocks in each of these DLCs are all actively traded and appear in
benchmark market indices, suggesting that liquidity differences are unlikely to be
15
able to explain substantial price differences.12 Furthermore, tax factors do not
appear to be able to justify the differential, since investors from third countries do
not obtain any major tax advantage from investing in a particular twin.13
Given that the ongoing large and variable price spreads in both the Anglo-Dutch
and Anglo-Australian DLCs are difficult to explain based on fundamentals, they
represent something of an anomaly. The reason why the price spread has not been
eliminated by investors is presumably related to the lack of fungibility (or
exchangeability) between the scrips. In particular, the different listings of a DLC
are distinctly different companies, with no fungibility that would allow
instantaneous riskless arbitrage.14 Instead, attempts by investors to take advantage
of the price differential via a long position in the discount stock and a short
position in the premium stock require investors to have long horizons and to be
able to withstand short-term losses should the differential widen.15 Indeed, one of
the positions held by Long-Term Capital Management (LTCM) during its 1998
collapse was a US$2.3 billion position in Royal Dutch/Shell that incurred losses
when the price differential widened sharply (see Lowenstein (2000)). This
12
The three Australian stocks are all included in the ASX 100 index, while the three UK stocks
have all been included in the FTSE 100 index, except that Brambles PLC slipped from the
FTSE 100 to the FTSE 250 in December 2002, in the last month of the sample studied here.
13 For example, in the Anglo-Australian cases, Australian investors would prefer the local stock
of a DLC due to the dividend taxation credits they receive (and which are not available to UK
investors). However, for third-country investors who can be viewed as the marginal investors,
the choice would depend on the respective tax treatment between their country and the
country of listing. In many instances, including the US case, current tax treaties suggest that
in most cases third-country investors would be largely indifferent between the Australian and
UK stock.
14 By contrast, price divergences between DLC twins and their respective ADRs in the New
York market are far smaller. For example, both Rio Tinto and BHP Billiton have ADRs
attached to each of their listings, and the prices of each ADR is invariably almost exactly
equal to the price in the respective primary listing. Since investors can exchange ADRs for the
underlying stock (i.e., the different scrip are fully fungible), price differentials can be
arbitraged away and are limited to the very small transactions costs involved in such
exchanges.
15 The lack of fungibility between the shares exposes the investors to noise trader risk, i.e. the
risk that the mispricing worsens in the short run (see De Long, Shleifer, Summers and
Waldmann (1990); Shleifer and Vishny (1997)). Abreu and Brunnermeier (2002) propose an
additional related risk, which they refer to as synchronisation risk, and is based on the
uncertainty an arbitrageur faces about when his or her peers will exploit an arbitrage
opportunity.
16
illustrates the difficulty that investors face in attempting to arbitrage away such
premiums.
A less complete form of arbitrage would involve long-only investors switching out
of the premium stock and into the discount stock. However, this type of activity
may be limited by home biases or by mandates that limit the asset allocation of
institutional investors. For instance, a manager with a mandate to invest only in
Australian (UK) stocks may be prevented from purchasing the UK (Australian)
scrip, even though the two scrips offer the same set of dividends and have highly
correlated returns. An example of the existence of these home biases or mandate
restrictions can be seen in the share registry of the BHP Billiton twins, where the
majority of the largest 20 shareholders in BHP Billiton Ltd in late 2002 were
Australian institutions, although none of the top 20 shareholders in BHP Billiton
PLC were Australian institutions.16
One interesting attempt to circumvent the constraints imposed by country-based
mandates was the introduction by an Australian broker of warrants, traded on the
Australian Stock Exchange on the UK arms of BHP Billiton and Rio Tinto.17 The
intention was to provide Australian managers with mandates to invest only in
Australian securities with a means of effectively investing in foreign assets.
However, the low volume of trading in these warrants suggests limited arbitrage
activity of this nature.
Of course, switching between DLC twins should be more feasible in the case of
large institutional investors with a global mandate. Indeed, newspaper reports
suggest that there are instances of institutional investors switching between DLC
twins. For example, the US-based Capital Group, which had been the largest
investor in BHP Billiton Ltd, was reported to have switched in 2002 into BHP
Billiton PLC.18 However, the implication of the substantial price differences that
are observed on an ongoing basis is that this type of behaviour is fairly limited.
In principle, the data for the average price premia for these six DLCs, plus the
average premia for other cases prior to their unification, might offer some hope for
16
This information on stockholdings is taken from Bloomberg.
See ‘Offshore exposure with a little bit of DLC’, Australian Financial Review, 15 April 2002,
p 23.
18 See ‘Easy 10pc on BHP arbitrage’, Sydney Morning Herald, 28 August 2002, p 24.
17
17
understanding what factors are relevant in determining which (if any) company
will trade at a premium. Some preliminary analysis failed to offer any systematic
relationship, so we conclude that the sample of DLCs is too small to differentiate
between explanations for the direction of the premia. However, one interesting
observation comes from the Anglo-Australian DLCs, where the Australian twin
has traded consistently at a substantial premium in two of the three cases. This
specific evidence about how different national markets value the same set of equity
cash flows can be viewed as refuting claims that are sometimes made that
companies listed on smaller markets are at an inevitable cost-of-capital
disadvantage and can increase shareholder value by simply shifting their primary
listing to larger overseas markets.
4.
Data
In this section, we briefly outline the sources of data used in the empirical tests in
the remainder of the paper. We calculate returns for the various companies
involved in the tests using price and dividend data from Datastream or Bloomberg.
Data for market indices were obtained from Datastream and include series for the
Australian ASX 100, Belgian BEL 20, Dutch AEX, Finnish HEX 25, French
CAC 40, Swedish OMX, Swiss SPI, UK FTSE 100, and US S&P 500. Returns for
individual stocks and market indices were calculated in log-differenced form.
Exchange rate data for the Anglo-Australian DLCs are log changes based on
Sydney closing rates from Bloomberg while exchange rate changes for all other
DLCs are based on London closing rates from Datastream.
5.
Testing for Excess Comovement
The phenomenon of assets with similar cash flows trading at quite different prices
is interesting. While these differences do not provide an opportunity for pure
riskless arbitrage, they are nonetheless an anomaly. Froot and Dabora (1999) find
that the price differential is correlated with the relative performance of the markets
in which the twins trade most. They propose that this comovement with the market
index where most of a twin’s trading occurs is a reflection of prices in each market
being influenced by market sentiment.
18
Given the recent creation of three Anglo-Australian DLCs, this section tests
whether the twins in these companies are also subject to excess comovement.
Further, we also test whether the excess comovement for the longer-standing
Anglo-Dutch DLCs has endured. In each case we apply Froot and Dabora’s
methodology and test the hypothesis that each twin’s price comoves excessively
with the market on which it trades most. For example, comovement would imply
that positive market shocks in Australia are associated with an increase in the price
of the Australian twin relative to the price of the UK twin.19 Accordingly, we
regress the return differential between the two companies on the returns of the
Australian and UK market indices plus the currency return:
ria,t − rib,t = β 0 + β1rma,t + β 2 rmb ,t + δriexr
,t + ε t
(1)
where ria,t and rib,t are log returns on the Australian and UK twins in DLC i, rma,t and
rmb,t are the log returns on the Australian and UK stock market, and riexr
,t is the log
return on the AUD/GBP exchange rate. We also run the corresponding regression
for the Anglo-Dutch DLCs.
The use of standard market indices in such regressions potentially creates a bias
when one of the companies is included in a market index. For example, each of the
Australian and UK twins are included in the ASX 100 and FTSE 100, respectively.
However, similar to the evidence in Froot and Dabora (1999), we find the effect of
this bias is minor since each firm bears a relatively small weight in its index.
However, in the case of the tests for the Anglo-Dutch DLCs, when we use the
Dutch AEX 100 index we remove Royal Dutch due to its significant weight.
The null hypothesis is that changes in the price differential are uncorrelated with
the performance of the two national markets on which the DLCs trade, but may be
correlated with exchange rate movements since the dependent variable is the
difference between price changes of assets traded in different currencies. Indeed, if
the twins’ prices (in a common currency) always moved together, we would expect
a coefficient of minus unity on the exchange rate variable. For instance, an
appreciation in the Australian dollar/pound sterling exchange rate should result in a
19
For all three Australian-UK DLCs, the Australian and UK arms trade mostly on the
Australian Stock Exchange and London Stock Exchange, respectively.
19
relative increase (decrease) in the local-currency price of the UK (Australian)
scrip.
The alternative hypothesis based on the comovement phenomenon is that 1 will be
positive and 2 will be negative. For example, a shock to the overall Australian
(UK) equity market is expected to be associated with an increase (decrease) in the
local currency price of the Australian twin relative to the local currency price of
the UK twin. The implication is that the price differential is being driven to an
extent by market-specific liquidity shocks or relative market sentiment.
We have estimated Equation (1) using a wide range of return horizon, but for
brevity, Table 3 presents results only for horizons of 2, 5, 10 and 20 days. The
reason for presenting longer horizon returns is to ensure that any estimated excess
comovement is not due to asynchronous trading effects or very short-term liquidity
shocks, and also to get a sense of the persistence of such effects.20 These results all
involve rolling regressions using overlapping data, so the statistical tests are based
on Newey-West standard errors to account for the moving average error process
that is introduced.21
20
In the case of the Anglo-Australian DLCs there is typically an eleven-hour difference between
UK and Australian market closing times. One possible approach to minimise the problems of
asynchronous trading would be to use Australian closing prices and UK opening prices
(approximately a three-hour time difference). Unfortunately, data problems with opening
prices precluded this approach. Instead, we conduct the regressions under the implicit
assumption that most global equity price discovery occurs during Northern Hemisphere
trading hours, so that a majority of price discovery for the Anglo-Australian DLCs also occurs
during London trading. Hence, the regressions actually use Australian returns to day t and UK
returns to day t-1. However, the results from an estimation using day t returns for both
markets are similar, especially for longer-horizon returns.
21 Although the standard errors are corrected by the Newey-West procedure, there is no similar
adjustment to the adjusted R2, which is biased upwards, especially as the extent of overlap
increases.
Notes:
(0.08)
0.49
1 806
(0.09)
0.22
389
(0.16)
0.28
386
(0.18)
0.43
381
(0.24)
0.66
381
(0.12)
0.06
364
(0.16)
–0.01
(0.19)
0.19
361
(0.11)
–0.04
(0.24)
0.32
356
(0.11)
–0.04
(0.23)
0.49
356
(0.10)
–0.06
This table provides tests of whether the three recent Anglo-Australian DLCs also demonstrate the excess comovement phenomenon identified by Froot and
a
b
Dabora (1999). It reports estimates of Equation (1): ria,t − rib,t = β 0 + β1rma ,t + β 2 rmb ,t + δriexr
,t + ε t , where ri ,t and ri ,t are the log returns on the Australian and
UK twins, respectively, rma ,t and rmb ,t are the log returns on the Australian and UK stock markets, respectively, riexr
,t is the log return on the AUD/GBP
exchange rate, and all log-difference returns are multiplied by 100. To take account of asynchronous trading, the Australian returns are measured to day t and
the UK returns are measured to day t-1. Newey-West standard errors are shown in parentheses. Rejections of the null hypothesis at the 10, 5 and 1 per cent
levels are denoted by *, ** and ***, respectively. The sample period is January 1996–December 2002 for Rio Tinto, July 2001–December 2002 for BHP
Billiton, and August 2001–December 2002 for Brambles Industries.
(0.13)
0.37
1 816
(0.06)
–0.39
(0.08)
0.27
1 821
(0.13)
–0.03
(0.04)
0.22
Adjusted R
Observations 1 824
2
(0.12)
–0.02
β0
(0.03)
–0.15**
(0.09)
–0.08
(0.06)
–0.05
(0.06)
–0.02
(0.06)
–0.01
(0.07)
–1.07
δ
(0.07)
–0.97
(0.06)
–1.05
(0.11)
–0.73*
(0.07)
(0.04)
–0.76** –0.91
20-day
(0.04)
(0.04)
(0.04)
(0.02)
(0.08)
(0.08)
–0.35*** –0.56*** –0.71*** –0.85*** –0.71** –0.86
10-day
β2
5-day
0.65*** 0.29**
2-day
(0.11)
–0.07
0.42***
20-day
(0.06)
(0.06)
(0.06)
(0.03)
(0.14)
(0.14)
(0.11)
(0.07)
(0.19)
(0.14)
(0.11)
–0.55*** –0.45*** –0.38*** –0.34*** –0.40*** –0.23*** –0.22*** –0.29*** –0.33*** –0.22*** –0.15**
10-day
0.02
5-day
0.77*** 0.41*** 0.28**
2-day
0.24**
20-day
0.45*** 0.41*** 0.42***
10-day
Brambles
0.65***
5-day
BHP Billiton
β1
2-day
Rio Tinto
Table 3: Testing for Excess Comovement, Anglo-Australian DLCs
20
21
Our results confirm the excess comovement findings of Froot and Dabora. Almost
all coefficients are significantly different from zero at the 1 per cent level and of
the expected sign. At the 2-day horizon, it may not be surprising that 1 is strongly
positive and that 2 is strongly negative, i.e., that the relative price of the twins is
very substantially affected by the relative performance of their national markets.
However, any short-term effects from liquidity shocks should be largely absent in
longer-term returns. Yet at the 10-day horizon, for example, the estimates for 1
and 2 of around 0.30 and –0.25, respectively, imply that a 10 per cent increase in
the Australian (UK) benchmark index is associated with an increase in the relative
price of the Australian (UK) twin of around 3 (2½) per cent.
Furthermore, it may not be surprising that the exchange rate coefficient is
typically substantially less than –1 at the 2-day horizon, so that exchange rate
changes in the very short-term have significant effects on the price differential.
However, at longer horizons it also remains less than –1. For example, estimates
for 3 of around –0.8 at the 10-day horizon imply that a 10 per cent change in the
Australian dollar/pound sterling exchange rate would tend to be associated with a
2 per cent increase in the (common currency) relative price of the twin from the
country that has seen an appreciation.22
The observation that the magnitude of the coefficients on the two market indices
falls (and that on the exchange rate coefficient rises) as the return horizon
lengthens, suggests that excess comovement may be largely a short-run
phenomenon. However, results from regressions using 50-day returns (available
upon request) continue to show excess comovement for Rio Tinto and
BHP Billiton, suggesting that the comovement is still present at fairly long
horizons.
Given that Froot and Dabora’s finding of excess comovement was based on data
only up to 1995, it may be of interest to see if this phenomenon has continued more
recently. Accordingly, we estimated versions of Equation (1) for Royal
Dutch/Shell, Unilever and Reed Elsevier for 1996–2002, and also for 1989–1995,
the previous seven-year period. In the case of the first two DLCs, we follow Froot
and Dabora and include the return on the S&P 500 index and the change in the US
22
When we impose the restriction that 3 is equal to –1, the finding that
significantly different to zero remains robust.
1
and
2
are
22
dollar/pound sterling exchange rate to reflect the significant share of trading of the
Dutch twin that occurs in the US market. In all three cases, the dependent variable
is defined as the return on the Dutch twin less the return on the UK twin, so the
alternate hypothesis of excess comovement would suggest that the sum of the
coefficients on the Dutch and US market indices will be positive, and the sign of
the coefficient on the UK market index will be negative.
The results for the Anglo-Dutch DLCs tell a similar story to the tests for the
Anglo-Australian DLCs. Table 4 shows the results from the 10-day return
specification (results for other horizons are available from the authors). In almost
all cases, we reject the null hypothesis concerning the market indices at the
1 per cent level. Comparing the results of the two periods, there is some indication
that the degree of excess comovement has lessened somewhat since the end of
Froot and Dabora’s sample. However, we hesitate to cite this result as evidence of
a reduction in the degree of the anomalous return behaviour of these DLCs, since
the earlier data on the mean absolute price differential in Table 2 would not really
support such a conclusion.
The results for both the Anglo-Dutch and Anglo-Australian DLCs therefore imply
that excess comovement with aggregate market indices is a pervasive feature of the
pricing of DLCs. In addition, the results consistently suggest a less-than-unit
response of the relative valuation of the twins to movements in the exchange rate.
In a sense, it appears that the market pays too much attention to one type of
information (overall market indices) that should be irrelevant to the relative
valuation of the twins but pays too little attention to information (exchange rates)
that is relevant to their relative valuation.
23
Table 4: Testing for Excess Comovement, Anglo-Dutch DLCs
Company
Constant
AEX
FTSE
Sample period: January 1989–December 1995
Royal Dutch Shell 0.03
0.15*** –0.33***
(0.08)
(0.04)
(0.03)
Unilever
–0.01
0.19*** –0.41***
(0.10)
(0.04)
(0.04)
(a)
Reed Elsevier
0.37*** 0.26*** –0.53***
(0.07)
(0.04)
(0.04)
Sample period: January 1996–December 2002
Royal Dutch Shell –0.07
0.18*** –0.27***
(0.08)
(0.03)
(0.04)
Unilever
–0.06
0.14*** –0.21***
(0.10)
(0.05)
(0.06)
Reed Elsevier
–0.10
0.14*** –0.18***
(0.09)
(0.04)
(0.05)
S&P
EUR/GBP EUR/USD Adjusted R2
0.10***
(0.04)
0.03
(0.05)
–
–0.67
(0.06)
–0.54
(0.07)
–0.75***
(0.05)
–0.12
(0.04)
–0.15
(0.05)
–
0.40
0.00
(0.03)
0.10***
(0.04)
–
–0.67
(0.07)
–0.82
(0.09)
–0.78***
(0.06)
–0.09
(0.06)
0.02
(0.07)
–
0.40
0.28
0.39
0.25
0.28
Notes:
This table provides tests of whether the excess comovement phenomenon identified by Froot and Dabora
(1999) has persisted beyond their sample period. It uses 10-day returns and shows estimates from the
1
2
equation: ria,t − rib,t = β 0 + β1rma,t + β 2 rmb ,t + β 3 rmc ,t + δriexr
+ δriexr
+ ε t where: ria,t and rib,t are the log
,t
,t
a
b
c
returns on the Dutch and UK twins, respectively; rm ,t , rm ,t and rm ,t are the log returns on the Dutch, UK
1
2
and US stock markets, respectively; riexr
and riexr
are the log returns on the EUR/GBP and EUR/USD
,t
,t
exchange rates, respectively; and all log-differenced returns are multiplied by 100. In the case of the
Dutch exchange rate, the guilder is used in place of the euro prior to 1999. Newey-West standard errors
are shown in parentheses. Rejections of the null hypothesis at the 10, 5 and 1 per cent levels are denoted
by *, ** and ***, respectively, except in the case of the regression coefficients for the exchange rate
variables in the Royal Dutch/Shell and Unilever equations, where there is no prediction about the size of
the parameter estimates.
(a) Sample period is January 1992–December 1995.
6.
Testing for Changes in Market Exposures Following
Unification of DLCs
We next explore another aspect of the excess comovement phenomenon based on a
different group of DLCs. In particular, we exploit the fact that six DLC structures
have been discontinued within the last decade, and replaced with a more
conventional single company structure. These firms are listed in Table 5, and some
further details of the unification of their share structures are provided in Section 7.
September 1998–
October 2000
November 1996–
February 2000
December 1997–
March 2000
January 1988–
July 1999
June 1990–
December 2001
July 1989–
April 1996
Period as a DLC
25 500
19 200
£19 327
28/08/2000
20/02/1996
2.5
0.2
12.5
4.5
7 257
4 475
20/09/1999
US$10 800
US$10 000
9.5
5 233
4 819
20/09/1999
04/02/1999
11.7
£10 908
£17 309
t=–3
2.1
0.4
6.1
0.0
0.2
6.0
t=+1
Per cent
Premium
17/04/2000
t=0
Market
Unification
announcement capitalisation
Million
date
In each case, the company trading at a discount prior to unification is listed second. Market capitalisation data were obtained from Bloomberg and company
annual reports as close as possible to the unification date (t=0).
Insurance and
financial services
Banking and
financial services
Banking and
financial services
Electrical engineering
and infrastructure
Banking and
financial services
Pharmaceuticals
Zurich Allied (Switzerland)
Allied Zurich (UK)
Dexia (Belgium)
Dexia (France)
Nordbanken (Sweden)
Merita (Finland)
ABB AG (Switzerland)
ABB AB (Sweden)
Fortis (B) (Belgium)
Fortis (NV) NL (Netherlands)
SmithKline Beecham (USA)
SmithKline Beecham PLC (UK)
Notes:
Industry
Company and country
Table 5: Details of Unified DLCs
24
25
In cases where two companies become one, the listing arrangements typically
change from two primary listings to a single primary listing in one market and a
secondary listing in the other. In some cases the company may remain in the
national market index of the country that is now the secondary listing, but
international index providers such as Morgan Stanley Capital International
(MSCI), Dow Jones and Bloomberg typically transfer the entire index weight to
the country of the new primary listing. This is likely to have implications for the
location of the trading of the company’s shares, with a shift towards the market of
the new single primary listing. However, if markets are perfectly integrated and the
fundamentals, or firm-specific risks, associated with the combined company have
not changed then the exposures of the combined firm to various markets should not
have changed.
However, studies have shown that location of trade and the way that stocks are
traded can influence stock prices. In addition to Froot and Dabora (1999), Chan,
Hameed and Lau (2003) investigate the case of the Jardine Group companies that
shifted their listings from Hong Kong to Singapore in 1994. They find that the
stocks became more correlated with the Singapore market after the change in
listing, despite the core business of the group remaining in Hong Kong. Two other
related examples are studied by Barberis, Shleifer and Wurgler (2002) and Sosner
and Greenwood (2002), who examine changes in correlation following inclusions
or exclusions from major stock indices, namely the US S&P 500 index and the
Japanese Nikkei 225 index, respectively. Both sets of authors find that when a
stock is added to an index it becomes substantially more correlated with the other
stocks in that index, and less correlated with stocks that are not in the index, with
an equivalent impact for deletions from an index. The unification of a DLC
provides another opportunity to examine how stock prices are affected by tradingrelated factors.
The prediction of a model of trading-induced comovement is that following the
unification of a DLC and the shift to a single primary listing on one market, the
exposure of the market value of the combined group to the market index of the
market of the new single primary listing will rise. Similarly, there should be a fall
in the market exposure of the combined group to the market that no longer has a
primary listing. That is, just as the existence of a DLC arrangement allows for
predictions about comovement, so to does the unification of a DLC arrangement
into a conventional merger.
26
We test the prediction of trading induced comovement by estimating a model
similar to that estimated by Chan, Hameed and Lau (2003) and focusing on four
unifications where there are clear predictions about market exposures. We regress
the return of the combined company, both before and after unification, on the
returns of the market indices of the two markets where the DLC traded. The return
on the combined company post-unification is directly observed from the trading of
the company on the new primary or new secondary market. However, we do not
directly observe the return on the combined company in the pre-unification period.
Hence, we calculate it as:
(
ricomb
= αria, t + (1 − α ) rib,t − riexr
,t
,t
)
(2)
where α is the weight of the twin (company a) in the market that becomes the new
primary listing, 1-α is the weight of the twin (company b) that becomes the new
secondary market listing, ria,t and rib,t are the returns of the respective companies,
and riexr
,t is the change in the exchange rate between the two countries in question.
Although the problem of differences in trading hours is not a significant issue for
the DLCs that have unified, we take account of possible non-trading effects or
asynchronicity of trading by estimating market exposures using 2-day returns
(though the results are not especially sensitive to this).23 We test for a change in
market exposures by estimating the following equation:
ricomb
= β1 + β 2 rmp, t + β 3rms , t + β1* Dt + β 2* Dt rmp,t + β 3* Dt rms , t + ε i ,t
,t
(3)
where rmp,t and rms ,t are the log returns of the market indices for the new primary
and new secondary market indices, respectively, and Dt is a dummy variable taking
the value zero in the pre-unification period and unity in the post-unification
period.24 We estimate this regression using data for 200 days prior to the
23
All are cases of European firms with no more than one-hour’s difference in trading hours,
with the exception of SmithKline Beecham which traded in the US and UK and was subject to
a five-hour difference in trading hours.
24 For presentational purposes, the model and results we present exclude the change in the
exchange rate as an additional explanatory variable. The results are, however, broadly similar
if this is also included.
27
announcement of unification, and 200 days post-unification beginning 20 days
after the first trading day as a unified company.25
The parameter estimate on the dummy interaction terms provides the test for
whether there has been a significant shift in market exposures. The null hypothesis
that the market exposures of the combined company are not affected by the
changed trading arrangements is that these dummy interaction terms are zero. The
alternate hypothesis of a change in market exposures due to changed trading
arrangements is that β 2* will be positive and β 3* will be negative.
Although we have data for six unified DLCs, we omit two of these in conducting
the tests for changes in betas. In the case of Merita Nordbanken, the FinnishSwedish financial group (which was renamed Nordea following its unification), the
announcement of the unification of the share structure was accompanied by the
announcement of a takeover of Norway’s Christiania Bank. We omit this case
since the takeover could have resulted in a fundamentals-based change in the
market exposures of the combined company. We also omit the case of the BelgianDutch Fortis financial group because the unification was accompanied by the
announcement that the group would retain dual primary listings. Indeed, different
providers of regional and global indices have taken different decisions as to which
of the two listings is included in their indices.
This leaves four cases where the unification of the share was not accompanied by
any changes in the underlying business of the group and where the unification
announcement explicitly stated that one particular market would become the new
primary market for the group. These include: ABB, the Swiss/Swedish industrial
conglomerate; Dexia, the Belgian/French financial firm; SmithKline Beecham, the
Anglo-American pharmaceutical company which has subsequently merged with
Glaxo Wellcome to form GlaxoSmithKline; and Zurich Allied/Allied Zurich, the
Swiss/UK insurer which is now called Zurich Financial Services.26 In these cases,
although the merged company remained in the market index of each country,27 the
25
By combining the pre- and post-unification data into a single regression, we are implicitly
making the assumption that the variance of the residual is equal in the two different periods,
an assumption which is not rejected by Chow tests.
26 The post-unification primary market is listed first where applicable.
27 An exception is Allied Zurich, which was deleted from the FTSE 100.
28
combined group is treated by all major global index providers (including MSCI,
FTSE, Bloomberg, and Dow Jones) as belonging entirely to one market.
The results for the tests for changes in market exposures following DLC
unification are shown in Table 6. The results provide very clear evidence that the
betas of the combined firm change very markedly following the unification of the
share structure. Furthermore, the results are exactly as would be predicted by a
model of trading-induced comovement. In particular, the beta for the market that
becomes the single primary listing for the merged company rises in all cases by
about 0.4 and there is a corresponding fall in the beta for the market that is now the
secondary listing. All the estimated changes are significant at the 10 per cent level,
except for the estimate for β 3* in the SmithKline Beecham regression, where the
result appears to reflect an atypical low parameter on the US market in the
particular pre-unification sample that was used.
Table 6: Testing for Changing Market Exposures in Unified DLCs
ABB
Dexia
Zurich
SmithKline
Beecham
Notes:
2
β1
β2
β3
β 1*
β 2*
–0.30*
(0.17)
–0.07
(0.12)
–0.25
(0.17)
0.97***
(0.10)
0.31***
(0.11)
1.19***
(0.17)
0.34***
(0.08)
0.38***
(0.09)
0.26*
(0.14)
0.45*
(0.24)
0.20
(0.17)
0.06
(0.24)
0.34**
(0.17)
0.28**
(0.13)
0.53**
(0.25)
–0.34***
(0.11)
–0.36***
(0.11)
–0.39*
(0.20)
0.55
0.19*
0.76***
0.03
–0.09
0.56***
–0.04
0.26
(0.11)
(0.14)
(0.15)
(0.15)
(0.20)
(0.19)
β 3*
Adjusted R
0.31
0.44
This table provides tests for whether the betas or market exposures of the combined DLCs change
following the unification of the share structure into a single company. It uses 2-day returns and shows
estimates from Equation (3): ricomb
= β1 + β 2 rmp,t + β 3rms ,t + β1* Dt + β 2* Dt rmp,t + β 3* Dt rms ,t + ε i ,t where
,t
ricomb
= αria,t + (1 − α ) rib,t − riexr
is the return of the combined DLC. α represents the pre-unification
,t
,t
weight of the primary stock. After unification α = 1. rmp, t and rms , t are the returns of the primary and
secondary markets, respectively. Dt is a dummy variable that is equal to 1 in the post-unification period
and 0, otherwise. The returns used are 2-day rolling log returns, with log-differences multiplied by 100.
Newey-West standard errors are shown in parentheses. Rejections of the null hypothesis at the 10, 5 and
1 per cent levels are denoted by *, ** and ***, respectively.
(
)
29
One possible explanation for the change in market exposures might be that the
value of the unified firm is measured solely on the market that is now the primary
listing, whereas the value of the pre-unification combined firm is measured partly
in the market which is now the secondary market listing. Hence, it could
potentially be the case that the shift in betas is due to the change in the source of
prices for the pre- and post-unification returns. This possibility was examined by
estimating Equation (3) with post-unification returns calculated from market prices
on the secondary market (and changing the currency of measurement for the preunification returns.) The results of this alternate test are almost identical to the
results shown in Table 6, which is not surprising since the prices of the unified
company are essentially equal on the primary and secondary markets.
Hence, we conclude that even though there is no change in the fundamental
exposures of these companies, the conversion of a DLC arrangement into a
conventional merger results in substantial changes in exposures to the two markets
where the DLC twins previously traded. Since unification results in changes in the
twins’ weights in global and regional indices, this result corresponds closely to the
results of Barberis, Shleifer and Wurgler (2002) and Sosner and
Greenwood (2002), who find related results for index inclusions and deletions in
the United States and Japan. In addition, the shift from two primary listings to a
single primary listing is somewhat analogous to the relisting of the Jardine Group
companies studied by Chan, Hameed and Lau (2003), and our results are also very
similar. Overall, our results are exactly as would be predicted by a model of
trading-based comovement, whereby the pricing of assets is partly determined by
the location of trade and the investors who trade them.
7.
An Event Study of Announcements of DLC Unifications
In the final analysis of the paper, we conduct an event study into the performance
of DLC twins around the time of announcements that the DLC structure would be
discontinued and replaced by a more conventional single company structure. We
do so using data for the six DLC arrangements listed in Table 5 that chose to unify
their share structure companies over 1996–2001. Not surprisingly, the pricing of
30
the twins converges substantially in these cases,28 but it is of interest to ask if this
occurred via an increase in the value of the company that was trading at a discount,
or a fall in the share price of the twin that was trading at a premium. Alternatively,
both share prices might have risen, or both might have fallen. Indeed, any evidence
for systematic gains or losses in overall market value might provide some
indication of how investors view DLC arrangements relative to more conventional
unified share structures.
We define the day of the announcement that management is proposing to unify the
share structure as our event day (t=0), and denote ari ,jt as the abnormal return of
company j in DLC i for period t, which is given by:
ari ,jt = ri ,jt − E ( ri ,jt )
(4)
where E (ri ,jt ) is the expected log return from the market model.
We estimate the market model for each company as:
ri ,jt = α + β1ria, t + β 2 rib,t + β 3riexr
,t + ε t
(5)
where ria,t and rib,t are the log returns for markets a and b, and riexr
,t is the log change
of the relevant exchange rate for period t. The market model is estimated over a
100-day window ending 20 days prior to the announcement of unification. We then
tested whether announcement-window abnormal returns were significantly
different from zero, looking both at the abnormal return on announcement day, and
the cumulative abnormal return over a three-day period centred on the
announcement day.
We also investigate if the value of the combined firm changes around the
announcement of unification. We do this by replacing the individual company
return ri ,jt in Equations (4) and (5) with the company return ricomb
described in
,t
Equation (2).
28
In several cases, the convergence at the time of the announcement is not complete, which is
presumably due to the uncertainty as to whether shareholders will vote to approve
management’s plans to unify the share structure.
31
One problem with the event study for the impact of DLC unification
announcements is that four of the six announcements occurred in conjunction with
other potentially value-relevant announcements.29 These confounding events
weaken our ability to single out the impact of the unification announcement on
overall company value. However, these events should have no impact on the tests
for the relative performance of the twins. An additional problem is that in some
cases special payments were paid to shareholders in just one twin. Although these
may impact on the relative valuation of the twins, they have no impact on the
overall company value.30
The results of the event study are summarised in Table 7, which shows the excess
returns for the six company pairs for both the announcement day and the 3-day
announcement window. Three individual twins trading at a discount prior to the
announcement (ABB AB, Merita and Dexia France) showed large positive
abnormal returns (of 5 per cent or more) on the event day that were significant at
the 1 per cent level. In addition, two companies trading at a premium (Dexia
Belgium and Zurich Allied) showed significant negative abnormal returns (of
around 4 per cent) on the day of announcement. On average, the companies trading
at a discount rose by 3 per cent, while the premium companies fell by 1.6 per cent,
with both of these averages being significantly different to zero at the 1 per cent
level.
29
This was the case for ABB (annual profit was US$1.305 billion relative to average
expectation of US$1.23 billion), SmithKline Beecham (annual profit was £1.36 billion
relative to average expectation of £1.33 billion), and Fortis (half-year net income was
1.56 billion relative to average expectation of 1.54 billion). Merita Nordbanken made its
announcement in conjunction with the takeover of Christiania Bank. These data on earnings
expectations are taken from stories in Bloomberg.
30 There are payments in three such cases, mostly to shareholders in the twin losing the primary
listing, but in one case to the twin trading at a premium and gaining the primary listing.
32
Table 7: Excess Returns to DLC Twins around Unification Announcements
Company
(a)
ABB AB
ABB AG
Fortis (NL)(a)
Fortis (B)
Merita(a)
Nordbanken
SmithKline Beecham(a)
SmithKline Beecham PLC
Allied Zurich(a)
Zurich Allied
Dexia France(a)
Dexia Belgium
Average for discount companies
Average for premium companies
Notes:
Individual twin
Difference
between twins
Total
company value
1-day
3-day
1-day
3-day
1-day
3-day
8.2***
1.6
0.5
0.4
5.3***
–0.6
–1.2
–1.7*
0.1
–4.7***
5.0***
–4.8***
3.0***
–1.6***
14.5***
8.0**
–4.4
–3.3
6.3**
0.6
–2.9
–3.0*
0.0
–7.2***
8.3***
–1.1
3.6***
–1.0
6.6***
6.5**
4.8**
11.1***
0.4
–4.0
1.7
2.9
0.1
–1.1
5.9***
5.8**
0.5
0.1
–1.4*
–2.9*
4.9***
7.2**
–2.7*
–4.3*
9.8***
9.4***
–0.2
3.7**
4.6***
4.7***
0.8
1.9*
This table shows the excess returns (based on a market model) of DLC twins around the date of the
announcement of the unification of the DLC into a unified share structure. The 3-day returns are
cumulated from one day before the event day (the unification announcement) to one day after. The results
show log returns multiplied by 100. The statistics for the difference between the twins are for the
performance of the company trading at a discount relative to the company trading at a premium.
Significance at the 10, 5 and 1 per cent levels is denoted by *, ** and ***, respectively.
(a) Company previously trading at a discount.
In four cases we can reject the hypothesis that there was no difference in the
relative performance of the twins, and not surprisingly, in each case it was the
discount company that outperformed. The two cases where there is no significant
difference in the performance of the twins (SmithKline Beecham and Fortis) are
those DLCs with the smallest price differentials prior to the announcement. On
average, the return differential for the four DLCs with a significant
33
pre-announcement price differential (ABB, Merita Nordbanken, Allied Zurich, and
Dexia) was around 5 per cent.31
Over the 3-day period the average cumulative abnormal return of the discount
firms was 3.6 per cent, significant at the 1 per cent level. The average cumulative
abnormal return for the premium companies was –1 per cent, but not significantly
different to zero. Hence, the significance of the negative abnormal returns for the
firms trading at a premium is dependent upon the length of the return window
used.
In terms of total firm value, the average across all six cases was a small, but
statistically insignificant, increase on the day of the announcement. Only one
group, ABB, recorded a statistically significant positive return on the event day,
but two groups showed weakly significant falls in market value. Thus, the overall
picture on the announcement day is one of increases in the value of the twins
trading at a discount being mostly offset by falls in the value of the twins trading at
a discount. However, when the return horizon is increased to 3 days, we find an
increase in average overall firm value over the 3-day window that is significant at
the 10 per cent level. However, the problem of confounding events, mostly positive
in nature, means that we cannot be too categorical in attributing it to the unification
announcement.
At one level, the finding of modest gains should not be surprising given the way in
which the unification has occurred. Unification typically involves the company
announcing that its new single primary listing will be on the market that – at least
at the time of the announcement – placed the higher valuation on the cash flows of
the twin companies.32 That is, management might be considered to be undertaking
a form of arbitrage by closing out the dual listing on the market that attached a
31
Since this can be thought of as the excess return on a notional long-short position, it indicates
that – at least at times of unification announcements – arbitrage positions in DLCs can be
quite profitable. This suggests that it might be of interest to revisit the work of Rosenthal and
Young (1990) with a much larger sample of DLCs and see whether arbitrage trading would
have been profitable more generally (i.e., not just looking ex post at cases of unification).
Gatev, Goetzmann and Rouwenhorst (1999) provide an example of an empirical study of
‘pairs trading’ based on the identification of highly correlated equities.
32 The case of Fortis – where there was little discount or premium – is an exception, since the
single share structure that eventuated did not have a single primary listing but had a dual
primary listing on the Belgian and Amsterdam exchanges.
34
lower valuation and moving the entire listing to the market which valued the
company more highly. However, given that these differences in valuations on
different markets might be relatively transitory – as well as the confounding events
– the finding of possibly modest gains in overall firm value presumably cannot be
generalised to any strong conclusions about the way that markets value DLC
structures versus unified share structures.
8.
Conclusion
The results of this paper bolster the findings of Froot and Dabora (1999) by
showing that price divergences between DLC twins and excess comovement in
market valuation are pervasive phenomena in a larger sample of DLC cases. An
innovation of the paper has been to study cases where DLC structures are
abandoned in favour of a more conventional unified share structure. Even though
the fundamentals of the combined firm should not have changed, we find that the
market exposures of the combined firm do change, and in the direction that would
be implied by models of trading-based comovement. Our results are consistent
with recent work by Barberis, Shleifer and Wurgler (2002), Sosner and
Greenwood (2002), and Chan, Hameed and Lau (2003) who study how the
covariance of returns changes due to changes in the index composition or trading
location. Together with the excess comovement finding, these results all point to a
model where asset prices are determined by more than fundamentals and are
influenced by the location of trade, the investors who trade them, and the manner
by which trading occurs.
Our survey of DLC structures (shown in Table 1) and the recent evidence –
including the two Anglo-Australian DLCs formed in 2001, the Anglo-South
African Investec DLC implemented in 2002, and the DLC arrangements proposed
recently by both US bidders for UK-listed P&O Princess Cruises PLC – suggests
that there may well be further instances of the DLC structure as the process of
corporate international expansion continues.33 However, the survey also indicates
33
Hancock, Gray and Sommelet (2002) also suggest that DLC arrangements may become more
frequent and involve companies from a wider range of countries than has hitherto been seen.
35
that many of the DLC arrangements put in place over the last 15 years have been
replaced by more conventional unified share structures. Subject to the caveat of the
small sample and confounding events, our event study of unification
announcements suggests that there is little evidence that unifying a DLC – when
this occurs on the market that trades at a premium – has a substantial effect on
overall combined firm value. This result suggests that there are at least some cases
where markets see the choice between a DLC arrangement and a conventional
merger as having little impact on firm value. It is unlikely of course that markets
are always indifferent between the two structures, so most mergers will no doubt
continue to be conventional ones.
36
References
Abreu D and MK Brunnermeier (2002), ‘Synchronization risk and delayed
arbitrage’, Journal of Financial Economics, 66(2/3), pp 341–360.
Barberis N and R Thaler (2002), ‘A survey of behavioural finance’, NBER
Working Paper No 9222, forthcoming in Handbook of the Economics of Finance.
Barberis N, A Shleifer and J Wurgler (2002), ‘Comovement’, NBER Working
Paper No 8895.
Chan K, A Hameed and ST Lau (2003), ‘What if trading location is different
from business location? Evidence from the Jardine Group’, Journal of Finance,
forthcoming, June.
De Long JB, A Shleifer, L Summers and R Waldmann (1990), ‘Noise trader
risk in financial markets’, Journal of Political Economy, 98(4), pp 703–738.
Froot KA and EM Dabora (1999), ‘How are stock prices affected by the location
of trade?’, Journal of Financial Economics, 53(2), pp 189–216.
Gatev E, W Goetzmann and K Rouwenhorst (1999), ‘Pairs trading:
performance of a relative based arbitrage rule’, NBER Working Paper No 7032.
Glanz S and G Sanderson (2001), ‘Dual listed companies’, Mergers &
Acquisitions Practice Group Newsletter, Baker & McKenzie, July, Sydney.
Hancock S, M Gray and C Sommelet (2002), ‘Dual-headed structures revisited’,
Global Counsel, 7, October, pp 14–27.
Hancock S, B Phillips and M Gray (1999), ‘When two heads are better than
one’, European Counsel, June, pp 25–37.
Karolyi GA (1998), ‘Why do companies list shares abroad?: A survey of the
evidence and its managerial implications’, Financial Markets, Institutions &
Instruments, 7(1), pp 1–60.
37
Lee C, A Shleifer and R Thaler (1991), ‘Investor sentiment and the closed-end
funds puzzle’, Journal of Finance, 46, pp 75–109.
Lowenstein R (2000), When Genius Failed: The Rise and Fall of Long-Term
Capital Management, Random House, New York.
MacKinlay C (1997), ‘Event studies in economics and finance’, Journal of
Economic Literature, 35(1), pp 13–39.
Mullainathan S and R Thaler (2000), ‘Behavioral economics’, NBER Working
Paper No 7948.
Pagano M, AA Röell and J Zechner (2002), ‘The geography of equity listing:
why do companies list abroad?’, Journal of Finance, 57(6), pp 2651–2694.
Rosenthal L and C Young (1990), ‘The seemingly anomalous price behaviour of
Royal Dutch/Shell and Unilever N.V./PLC’, Journal of Financial Economics,
26(1), pp 123–141.
Shleifer A (2000), Inefficient Markets: An Introduction to Behavioral Finance,
Clarendon Lectures in Economics, Oxford University Press, New York.
Shleifer A and R Vishny (1997), ‘The limits of arbitrage’, Journal of Finance, 53,
pp 35–55.
Smith C and V Cugati (2001), ‘Innovative structures – dual listed companies’,
Focus on Mergers and Acquisitions, Allens Arthur Robinson, Sydney.
Sosner N and R Greenwood (2002), ‘Trade and the comovement of stock returns:
evidence from Japan’, unpublished paper, Harvard University.